According to Vuforia Engine™, the marker’s score was 4 in a 1–5 scale that rates the quality of markers for their optimal detection. App 1 allowed for the detection of a digital image marker (referred to as “marker” henceforth) by the headset’s camera. The camera resolution of this headset is 2.4 megapixels. For instance, the recommendation for the Vuforia Engine™ is a minimum width of image markers calculated by dividing the camera-to-image-marker distance by 10, which must be adjusted based on the resolution of the camera used.Īn AR app (App 1) for the holographic headset HoloLens ® (first generation) was created using Vuforia Engine™ (version 6.2.10) and Mixed Reality Toolkit (version 1.5.8.0).
High-resolution cameras allow the use of small image markers and/or their location far from the camera while preserving their optimal detection and correct registration of the virtual model. For example, previous research has demonstrated that the accuracy of image marker pose estimation is affected by the image marker size and camera-to-image-marker distance and angle. In addition, achieving an optimal balance between image marker size, camera-to-image-marker distance and camera resolution is key to maximise registration accuracy. edges or corners) which may reduce the positional accuracy with which virtual models are rendered. They attributed this effect to slight differences in the detection of image features (e.g. argued that detection of image markers becomes unstable when they lie on the central axis of the camera view frustum. The position and size of image markers may affect the image marker pose estimated by computer vision systems and thus the accuracy of the registration of virtual models with real-world features in AR applications. Brainlab™ or Medtronic StealthStation™ provide trajectory angle and positional errors of ≤ 2° and ≤ 2 mm, respectively. To use holographic headsets for surgical guidance, their accuracy must be equal or below that one of currently approved navigation systems, e.g.
However, image marker detection and rendering stability, and thus registration accuracy, may be affected by several factors, e.g. During surgery, these image markers are recognised by image-pattern-recognition tools that compute the registration (Fig. radio-opaque image markers) to the patient’s body at the time of scanning. Registration of virtual models to the patient’s body may be achieved by fixing fiducial markers (e.g. the correct trajectories of surgical instruments) to the operating room. This helps to transfer image data produced during the planning of the surgery (e.g. Emerging augmented reality (AR) technologies such as holographic headsets allow the overlay of patient-specific virtual models obtained from medical scans on the patient’s body surface in a predefined position.